Alper Nebi Kanlı

Data Science · LLM Engineering · AI Agents · Information Retrieval

I build with large language models and write about how they actually work: embeddings, retrieval, the architectures underneath. Beyond the profession I read widely in linguistics, philosophy, and the natural sciences, and that tends to leak into the writing.

interpretability
Reading and editing what language models represent, and what that means for safety.
refusal in llms is not one thingAugust 3, 2026
representational convergence
Essays on what it means for models to learn the same structure.
representational convergence is not one thing - part 2July 13, 2026representational convergence is not one thing - part 1May 20, 2026
models
Posts about specific machine learning models.
some berts and modernbertMay 22, 2025